Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Survival analysis in Python: Kaplan-Meier, Cox proportional hazard, Aalen additive, parametric models, and competing risks. Censored data handling for churn, clinical, and actuarial applications.
.claude/skills/mkurman-lifelines/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 53% | 0% |
| case-13 | ✓→✓ | ＝ Same ✓ | 22% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -20% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -23% | 0% |
Lifelines is a survival analysis library for Python. It implements Kaplan-Meier, Cox Proportional Hazard, parametric models (Weibull, Log-Normal), and Aalen's additive model. Use it for time-to-event data in clinical trials, churn analysis, reliability engineering, and customer retention studies.
bashuv pip install lifelines
pythonfrom lifelines import KaplanMeierFitter import pandas as pd T = pd.Series([5, 10, 15, 20, 25, 30]) # durations E = pd.Series([1, 1, 0, 1, 0, 0]) # event observed? kmf = KaplanMeierFitter() kmf.fit(T, E) kmf.plot_survival_function() print(kmf.median_survival_time_)
pythonfrom lifelines import CoxPHFitter df = pd.DataFrame({ "duration": [5, 10, 15, 20, 25, 30], "event": [1, 1, 0, 1, 0, 0], "age": [45, 60, 55, 70, 50, 65], "treatment": [1, 0, 1, 0, 1, 0], }) cph = CoxPHFitter() cph.fit(df, duration_col="duration", event_col="event") cph.print_summary() cph.plot_partial_effects_on_outcome("treatment", [0, 1])
pythonfrom lifelines import WeibullAFTFitter wbf = WeibullAFTFitter() wbf.fit(df, duration_col="duration", event_col="event") wbf.print_summary()
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | pass→pass | 6,042 | 4,286 | -29% | 1 | 1 | 0% | 983 | 1,197 | +22% | 0 | 0 | — |
case-01 | fail→pass | 10,901 | 8,184 | -25% | 1 | 1 | 0% | 2,178 | 2,063 | -5% | 0 | 0 | — |
case-02 | pass→pass | 9,743 | 5,604 | -42% | 1 | 1 | 0% | 1,830 | 1,456 | -20% | 0 | 0 | — |
case-03 | pass→pass | 8,827 | 3,883 | -56% | 1 | 1 | 0% | 1,673 | 1,286 | -23% | 0 | 0 | — |
case-04 | pass→pass | 7,929 | 5,872 | -26% | 1 | 1 | 0% | 1,534 | 1,573 | +3% | 0 | 0 | — |
case-05 | pass→pass | 7,846 | 3,092 | -61% | 1 | 1 | 0% | 1,319 | 1,080 | -18% | 0 | 0 | — |
case-06 | pass→pass | 4,417 | 2,185 | -51% | 1 | 1 | 0% | 732 | 788 | +8% | 0 | 0 | — |
case-07 | pass→pass | 8,055 | 4,625 | -43% | 1 | 1 | 0% | 1,450 | 1,322 | -9% | 0 | 0 | — |
case-08 | fail→pass | 3,668 | 2,627 | -28% | 1 | 1 | 0% | 607 | 929 | +53% | 0 | 0 | — |
case-09 | pass→pass | 4,747 | 4,175 | -12% | 1 | 1 | 0% | 844 | 1,198 | +42% | 0 | 0 | — |
case-10 | pass→pass | 8,244 | 4,515 | -45% | 1 | 1 | 0% | 1,490 | 1,285 | -14% | 0 | 0 | — |
case-11 | pass→pass | 7,617 | 5,435 | -29% | 1 | 1 | 0% | 1,394 | 1,488 | +7% | 0 | 0 | — |
case-12 | pass→pass | 7,487 | 5,047 | -33% | 1 | 1 | 0% | 1,364 | 1,293 | -5% | 0 | 0 | — |
case-14 | pass→pass | 10,540 | 6,901 | -35% | 1 | 1 | 0% | 1,993 | 1,730 | -13% | 0 | 0 | — |
case-15 | pass→pass | 5,352 | 5,148 | -4% | 1 | 1 | 0% | 890 | 1,493 | +68% | 0 | 0 | — |
case-16 | pass→pass | 6,130 | 4,449 | -27% | 1 | 1 | 0% | 1,106 | 1,315 | +19% | 0 | 0 | — |
case-17 | pass→pass | 3,518 | 3,077 | -13% | 1 | 1 | 0% | 620 | 999 | +61% | 0 | 0 | — |
case-18 | pass→pass | 2,852 | 1,084 | -62% | 1 | 1 | 0% | 381 | 632 | +66% | 0 | 0 | — |
case-19 | pass→pass | 2,387 | 1,437 | -40% | 1 | 1 | 0% | 303 | 706 | +133% | 0 | 0 | — |
case-20 | pass→pass | 9,547 | 6,543 | -31% | 1 | 1 | 0% | 1,905 | 1,665 | -13% | 0 | 0 | — |
case-21 | pass→pass | 18,661 | 19,728 | +6% | 1 | 1 | 0% | 3,669 | 4,527 | +23% | 0 | 0 | — |
case-22 | pass→pass | 10,221 | 7,154 | -30% | 1 | 1 | 0% | 1,843 | 1,769 | -4% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.